AI summaries & chat

Optional AI that runs on your machine, or against a server you control.

What it does

• Summaries — a clean Markdown TL;DR, key points, and action items. Pick a style preset or write your own prompt.
• Living project overview — one editable document, organized by project, that the AI keeps current after each meeting while preserving your edits.
• Cross-meeting chat — ask questions about one meeting or all of them. Answers come only from your transcripts, and it says when something wasn't discussed rather than inventing.

Two ways to run it

Built-in local model

Runs entirely on your device. You choose a model — options include Qwen2.5, Qwen3, Gemma, Llama 3.2, and Phi-3.5 — which downloads once on first use (roughly 1.5 GB and up, depending on the model).

Bring your own server

Point Zerds at any OpenAI-compatible endpoint — Ollama, LM Studio, llama-server, vLLM — in Settings → LLM. Transcript text is sent only to the endpoint you configure; the bearer token is stored locally, and connections use your OS trust store.